gradient fails when output depends on multi-dimensional generator with a condition

[AwesomeQuest]

Gradient fails when calling with a function that depends on a 2D generator with a condition like the following

gradient([0.,0]) do pts
        N = length(pts)
        reduce(+, i for i in 1:N , j in 1:N if j>i)
end

The error does not happen when replacing the ',' with a "for"

gradient([0.,0]) do pts
        N = length(pts)
        reduce(+, i for i in 1:N for j in 1:N if j>i)
end

Full Stack-trace, package version, and julia version bellow

(@AD) pkg> st
Status `C:\Users\torfi\.julia\environments\AD\Project.toml`
  [e88e6eb3] Zygote v0.7.10

julia> versioninfo()
Julia Version 1.12.1
Commit ba1e628ee4 (2025-10-17 13:02 UTC)
Build Info:
  Official https://julialang.org release
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: 8 × Intel(R) Core(TM) i7-1065G7 CPU @ 1.30GHz
  WORD_SIZE: 64
  LLVM: libLLVM-18.1.7 (ORCJIT, icelake-client)
  GC: Built with stock GC
Threads: 1 default, 1 interactive, 1 GC (on 8 virtual cores)

julia> gradient([0.,0]) do pts
           N = length(pts)
           reduce(+, i for i in 1:N , j in 1:N if j>i)
       end

julia> show(err)
1-element ExceptionStack:
MethodError: no method matching length(::Nothing)
The function `length` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  length(::Tables.DictRowTable)
   @ Tables C:\Users\torfi\.julia\packages\Tables\cRTb7\src\dicts.jl:118
  length(::IRTools.Inner.IR)
   @ IRTools C:\Users\torfi\.julia\packages\IRTools\v0mn8\src\ir\ir.jl:208
  length(::Core.SimpleVector)
   @ Base essentials.jl:936
  ...

Stacktrace:
 [1] productfunc(xs::Tuple{UnitRange{Int64}, UnitRange{Int64}}, dy::Matrix{Union{Nothing, Tuple{Float64, Nothing}}})
   @ Zygote C:\Users\torfi\.julia\packages\Zygote\55SqB\src\lib\array.jl:278
 [2] (::Zygote.var"#product_pullback#product_pullback##0"{Tuple{UnitRange{Int64}, UnitRange{Int64}}})(dy::Matrix{Union{Nothing, Tuple{Float64, Nothing}}})
   @ Zygote C:\Users\torfi\.julia\packages\Zygote\55SqB\src\lib\array.jl:295
 [3] (::Zygote.var"#977#978"{Zygote.var"#product_pullback#product_pullback##0"{Tuple{UnitRange{Int64}, UnitRange{Int64}}}})(Δ::ChainRulesCore.InplaceableThunk{ChainRulesCore.Thunk{ChainRules.var"#thunked_∇getindex##2#thunked_∇getindex##3"{Matrix{Tuple{Int64, Int64}}, Vector{ChainRulesCore.Tangent{Any, Tuple{Float64, ChainRulesCore.ZeroTangent}}}, Tuple{Vector{CartesianIndex{2}}}}}, ChainRules.var"#thunked_∇getindex##0#thunked_∇getindex##1"{Matrix{Tuple{Int64, Int64}}, Vector{ChainRulesCore.Tangent{Any, Tuple{Float64, ChainRulesCore.ZeroTangent}}}, Tuple{Vector{CartesianIndex{2}}}}})
   @ Zygote C:\Users\torfi\.julia\packages\ZygoteRules\CkVIK\src\adjoint.jl:72
 [4] #49
   @ .\REPL[10]:3 [inlined]
 [5] (::Zygote.Pullback{Tuple{var"#49#50", Vector{Float64}}, Tuple{Zygote.ZBack{Returns{Tuple{ChainRulesCore.NoTangent, ChainRulesCore.NoTangent, ChainRulesCore.NoTangent}}}, ComposedFunction{Zygote.ZBack{ChainRules.var"#filter_pullback#rrule##125"{ChainRules.var"#getindex_pullback#rrule##252"{Matrix{Tuple{Int64, Int64}}, Tuple{Vector{CartesianIndex{2}}}, Tuple{ChainRulesCore.NoTangent}}}}, typeof(ZygoteRules.unthunk_tangent)}, Zygote.var"#405#406"{Zygote.Jnew{var"#51#52", Nothing, false}}, Zygote.Pullback{Tuple{Type{Base.Generator}, var"#51#52", Vector{Tuple{Int64, Int64}}}, Tuple{Zygote.var"#405#406"{Zygote.Jnew{Base.Generator{Vector{Tuple{Int64, Int64}}, var"#51#52"}, Nothing, false}}}}, Zygote.ZBack{Returns{Tuple{ChainRulesCore.NoTangent, ChainRulesCore.NoTangent}}}, Zygote.var"#231#232"{Zygote.var"#229#230"{Zygote.Context{false}, GlobalRef, Type{var"#51#52"}}}, Zygote.var"#405#406"{Zygote.Jnew{var"#53#54", Nothing, false}}, Zygote.var"#231#232"{Zygote.var"#229#230"{Zygote.Context{false}, GlobalRef, Type{var"#53#54"}}}, Zygote.ZBack{Returns{Tuple{ChainRulesCore.NoTangent, ChainRulesCore.NoTangent, ChainRulesCore.NoTangent}}}, Zygote.var"#977#978"{Zygote.var"#product_pullback#product_pullback##0"{Tuple{UnitRange{Int64}, UnitRange{Int64}}}}, Zygote.Pullback{Tuple{typeof(reduce), typeof(+), Base.Generator{Vector{Tuple{Int64, Int64}}, var"#51#52"}}, Tuple{Zygote.Pullback{Tuple{Type{NamedTuple}}, Tuple{}}, Zygote.Pullback{Tuple{Base.var"##reduce#276", @Kwargs{}, typeof(reduce), typeof(+), Base.Generator{Vector{Tuple{Int64, Int64}}, var"#51#52"}}, Any}, Zygote.var"#525#526"{Zygote.var"#pairs_namedtuple_pullback#pairs_namedtuple_pullback##0"{(), @NamedTuple{}}}}}}})(Δ::Int64)
   @ Zygote C:\Users\torfi\.julia\packages\Zygote\55SqB\src\compiler\interface2.jl:0
 [6] (::Zygote.var"#pullback##0#pullback##1"{Zygote.Pullback{Tuple{var"#49#50", Vector{Float64}}, Tuple{Zygote.ZBack{Returns{Tuple{ChainRulesCore.NoTangent, ChainRulesCore.NoTangent, ChainRulesCore.NoTangent}}}, ComposedFunction{Zygote.ZBack{ChainRules.var"#filter_pullback#rrule##125"{ChainRules.var"#getindex_pullback#rrule##252"{Matrix{Tuple{Int64, Int64}}, Tuple{Vector{CartesianIndex{2}}}, Tuple{ChainRulesCore.NoTangent}}}}, typeof(ZygoteRules.unthunk_tangent)}, Zygote.var"#405#406"{Zygote.Jnew{var"#51#52", Nothing, false}}, Zygote.Pullback{Tuple{Type{Base.Generator}, var"#51#52", Vector{Tuple{Int64, Int64}}}, Tuple{Zygote.var"#405#406"{Zygote.Jnew{Base.Generator{Vector{Tuple{Int64, Int64}}, var"#51#52"}, Nothing, false}}}}, Zygote.ZBack{Returns{Tuple{ChainRulesCore.NoTangent, ChainRulesCore.NoTangent}}}, Zygote.var"#231#232"{Zygote.var"#229#230"{Zygote.Context{false}, GlobalRef, Type{var"#51#52"}}}, Zygote.var"#405#406"{Zygote.Jnew{var"#53#54", Nothing, false}}, Zygote.var"#231#232"{Zygote.var"#229#230"{Zygote.Context{false}, GlobalRef, Type{var"#53#54"}}}, Zygote.ZBack{Returns{Tuple{ChainRulesCore.NoTangent, ChainRulesCore.NoTangent, ChainRulesCore.NoTangent}}}, Zygote.var"#977#978"{Zygote.var"#product_pullback#product_pullback##0"{Tuple{UnitRange{Int64}, UnitRange{Int64}}}}, Zygote.Pullback{Tuple{typeof(reduce), typeof(+), Base.Generator{Vector{Tuple{Int64, Int64}}, var"#51#52"}}, Tuple{Zygote.Pullback{Tuple{Type{NamedTuple}}, Tuple{}}, Zygote.Pullback{Tuple{Base.var"##reduce#276", @Kwargs{}, typeof(reduce), typeof(+), Base.Generator{Vector{Tuple{Int64, Int64}}, var"#51#52"}}, Any}, Zygote.var"#525#526"{Zygote.var"#pairs_namedtuple_pullback#pairs_namedtuple_pullback##0"{(), @NamedTuple{}}}}}}}})(Δ::Int64)
   @ Zygote C:\Users\torfi\.julia\packages\Zygote\55SqB\src\compiler\interface.jl:97
 [7] gradient(f::Function, args::Vector{Float64})
   @ Zygote C:\Users\torfi\.julia\packages\Zygote\55SqB\src\compiler\interface.jl:154
 [8] top-level scope
   @ REPL[10]:1